Re: Ann: SWI-Prolog 6.5.3

On Thu, Nov 21, 2013 at 4:47 AM, Jan Wielemaker <[email protected]> wrote:

> On 11/21/2013 05:47 AM, Michael Hendricks wrote:
>
>>
>>     In the same way, in the NU Prolog approach, a byte-per-element string
>>     *IS* a code list, just stored differently.
>>
>>
>> I think this is an important point worth extra attention.  In one sense,
>> [H|T] is just a comfortable interface for sequences of data.  A linked
>> list is the traditional sequence, but string-stored-as-byte-array is a
>> sequence too.  A sequence might be the key-value pairs in a map or the
>> bytes in a stream, etc.  Clojure sequences take the idea to its
>> conclusion by abstracting the cons interface so that many data
>> structures can implement it.
>>
>> I'd love to see this idea in more Prologs.
>>
>
> I don't think it is that easy.


I agree that it's not easy.  Prolog is a fantastic interface to my CPU's
instruction set, but it's taken a lot of hard work to get there :-)


>  [H|T] is syntactic sugar for a data structure, not for an operation.


We can also imagine it as sugar for a relation. foo([H|T]) might desugar to

    sequence(H,T,Whole),
    foo(Whole).

With coroutines and appropriate definitions for sequence/3 you can get both
reading and writing.  I've experimented with this arrangement on a couple
data structures.  I think it can extend to others, but I may be wrong.  If
it pans out, I'll release it as a pack.

-- 
Michael
-------------- next part --------------
HTML attachment scrubbed and removed